Precision Evaluation of Size-Selective Aerosol Samplers

Abstract

The importance of assessing exposure to atmospheric particles has recently increased, partially owing to epidemiological studies that have identified the negative health effects of particulate matter. Although size-selective sampling devices appear to be promising for measuring the degree of exposure to aerosol particles, the performance testing criteria of these devices may still be limited. Aerosol sampler performance can be measured in terms of bias and imprecision. The bias map, as a function of particle size distribution, is extensively used to evaluate sampling inaccuracy. However, procedures for constructing the imprecision map remain undetermined. The imprecision map should provide another useful indicator of sampler performance. This study develops a semi-empirical model of imprecision under statistical premises. The binomial distribution assumption, rather than the conventional normal assumption of an ANOVA test, was made to model the imprecision of the sampler. Analytical results indicated that the size distribution of challenge aerosols, total particle number count, number of specimens, and number of replicates all affected the imprecision map. The “One Standard Error Shift,” similar to the “Mean Square Error,” combining the bias and imprecision maps was a novel and effective indicator of sampler performance.